Fixed contract cost has a considerable influence on the aggregator’s profitability and can change its preferences with regard to portfolio composition. For example, due to a large number of contracts in Scenario 3, where the aggregator uses small home appliances’ flexi-bility, the level of fixed contract cost, naturally, has the biggest impact (see Figure 12 and Figure 13). When the cost level reachesd0,482 per year in Case 1 andd0,009 per week in Case 2, the savings in imbalance payments are outweighed by the fixed contract cost and the aggregator should revise it’s portfolio: the composition or the number of contracts (see Table 11).
Figure 12: The impact of fixed contract cost on imbalance payments reduction, Case 1 Figure 12 shows that in the longer run (Case 1 representing the whole year), when the fixed contract cost becomes relatively high, a portfolio of EVs (Scenario 1) provides the largest net benefit of using flexibility. Meanwhile, portfolios that include small home appliances
Figure 13: The impact of fixed contract cost on imbalance payments reduction, Case 2
(Scenario 3, 4, 6 and 7) are the least attractive, since the number of contracts and the total contract cost is large. In the short period (see Figure 13), situation is similar: with relatively high fixed contract cost the benefit of small home appliances’ flexibility cease to outweigh fixed contract cost sooner that in other portfolios that does not include this source.
However, in the analysed winter week, the number of heat pumps is the lowest comparing to other sources of flexibility, leading to lower total fixed contract cost and the highest net benefit of using their flexible load. In this case, fixed contract cost can be the highest among all portfolios, d0,104, and still sustain a non-negative profit from shifting the consumption of heat pumps (see Table 11).
In Figure 12 and Figure 13 we can distinguish contract cost intervals indicating the best portfolio choice for the aggregator. In Case 1, when the cost is relatively low,cc∈[0; 0,141), the aggregator would choose a portfolio with small home appliances’ flexibility. When the cost is in the interval cc ∈ (0,141; 0,184), the aggregator prefers a portfolio of mixed flexibility sources: small home appliances and EVs. And finally, when the cost gets relatively high, cc ∈ (0,184; 3,172), the portfolio of EVs is the most attractive. If the contract cost rises even higher, then aggregator should reduce the number of contracts in its portfolio and look for the most profitable option to shift flexible consumption. In Case 2, when the cost is in the intervalcc∈[0; 0,001), the aggregator prefers a portfolio of small home appliances, when cc∈(0,001; 0,038) – a portfolio of EVs, and whencc∈(0,038; 0,104) – a portfolio of
HPs.
The analysis suggests that under certain fixed contract cost levels a portfolio with mixed types of flexibility sources can be superior to the one with a single type of flexibility source.
However, in both investigated cases, most of the time the aggregator chooses a single type of flexibility source in which it would specialise depending on the fixed contract cost level.
Table 11: Maximum fixed contract cost to keep non-negative profit from shifting the con-sumption depending on a scenario and case, d
1 2 3 4 5 6 7
Case 1 3,172 2,897 0,481 0,939 3,035 0,762 0,736
Case 2 0,059 0,104 0,008 0,016 0,070 0,013 0,011
6 Conclusions and discussion
This paper focuses on the role of flexible demand aggregators and the effects of their portfolio choice on imbalance payments and compensations to flexibility providers. A game theoretical model has been used to simulate optimal flexible load schedules yielding the highest savings on imbalance payments. The model has five stages in which the consumer schedules his or her flexible load based on day-ahead prices, and the aggregator decides what flexibility prices would incentivise the consumer to shift flexible load and reduce payments for imbalance.
The effects of different portfolio compositions are reflected in seven scenarios. Nordic power market data for Denmark’s DK2 price area, as well as specific technical data for appliances and typical usage of appliances in Danish households allow to achieve realistic outcomes of demand-side flexibility employment for balancing purposes.
Results show that different compositions of flexibility sources (EVs, HPs, washing machines, clothes dryers and dish washers) influence the aggregator’s imbalance payments and com-pensations to consumers for provided flexibility. A portfolio of small home appliances seems to be the most attractive option for the aggregator. Moreover, it also yields the highest payoffs for the consumers. However, in all scenarios the compensation for the shifted load is too low to incentivise consumers to participate in flexibility trading. Consequently, ag-gregators might have to find other ways to encourage consumers to offer their flexibility, for
example, promote additional services.
An important factor in optimising consumption schedules is the length of the time interval for the flexible load shifting. Differences in regulating energy prices increase significantly in longer periods and higher volatility enables to yield higher savings on imbalance payments.
However, this is not a sufficient characteristic for the flexibility source to be the most effective. The effectiveness also depends on time when the flexibility is offered. Results indicate that flexibility offers during the day time are more valuable than at night. This is also related to the fact that price differences are larger during the day time.
With no fixed contract cost, there is no indication that portfolio diversification, in terms of different flexibility sources, could bring additional value in reducing imbalance payments.
Therefore, the aggregators might choose to specialise in certain types of flexibility sources that have a potential for maximum savings. However, if the aggregator incurs fixed contract cost, under certain cost level, the mix of different flexibility sources can be beneficial. Also, with increasing fixed contract cost, the aggregator might switch from one type of flexibility to another.
The difference between the forecasted and the actual reductions in imbalance payments is also affected by the portfolio composition. Portfolio of EVs allows to predict the reductions in imbalance payments with the smallest error, while the forecasted outcomes of HPs portfo-lio are the least accurate. In the latter scenario, frequent changes in consumption schedules lead to lower actual savings due to an increase in forecasting errors.
In a one-price balance settlement system the aggregator does not have to participate in the regulating energy market in order to benefit from excess demand flexibility. However, this model allows to determine potential flexibility trades in case of two-price balance settlement system. Due to the largest hourly shifts in load schedules, EVs’ flexibility can be traded the most. Furthermore, high value of minimum bid size, determined by current regulations, causes more difficulties for the aggregator to place bids of dispersed small home appliances flexibility than that of more fragmented but larger in size EVs’ flexibility. Thus, since diversification results in more dispersed flexibility offers, they are more likely to be used for imbalance payments reductions than excess flexibility trading.
There are several directions for future extensions of this research. In the future, increased renewable energy integration will change electricity prices and therefore potential value of demand-side flexibility. Modelling possible future scenarios would allow the aggregators to choose the best investment strategy and evaluate related risks. Aggregators could also be BRPs responsible not only for the consumption but also for the production side. Moreover, different portfolio compositions and introduction of penalties for not delivered flexibility could lead to different bidding strategies in the balancing market. This model is applied for the Nordic power market, thus, different geographical areas and peculiarities of other power markets could generate different outcomes. Also, the aggregator’s decisions could be influ-enced by its behaviour in the day-ahead and intraday markets. From the consumer’s point of view, different disutility functions or other contracts for changing flexible consumption schedule could also be investigated.
References
Agnetis, A., Dellino, G., Pascale, G. D., Innocenti, G., Pranzo, M. and Vicino, A. (2011),
‘Optimization models for consumer flexibility aggregation in smart grids: the ADDRESS approach’,2011 IEEE First International Workshop on Smart Grid Modeling and Simu-lation (SGMS) pp. 96–102.
Ali, M., Alahaivala, A., Malik, F., Humayun, M., Safdarian, A. and Lehtonen, M. (2015),
‘A market-oriented hierarchical framework for residential demand response’, Electrical Power and Energy Systems69, 257–263.
Biegel, B., Andersen, P., Stoustrup, J., Madsen, M. B., Hansen, L. H. and Rasmussen, L. H.
(2014a), Aggregation and Control of Flexible Consumers - A Real Life Demonstration, in
‘The 19th World Congress of the International Federation of Automatic Control’, Cape Town, South Africa.
Biegel, B., Westenholz, M., Hansen, L. H., Stoustrup, J., Andersen, P. and Harbo, S.
(2014b), ‘Integration of flexible consumers in the ancillary service markets’, Energy 67, 479–489.
Broberg, T. and Persson, L. (2016), ‘Is our everyday comfort for sale? Preferences for demand management’, Energy Economics54, 24–32.
Denmark’s Ministry of Transport (2012), ‘The Danish Transport System. Facts and Figure’.
Copenhagen, ISNB 978-87-91013-69-0.
Dong Energy (2013), ‘Alt om elforbrug’. No. 09.13.601-2013.
Energinet.dk (2008), ‘Regulation C2: The balancing market and balance settlement’.
Energinet.dk and Danish Energy Agency (2012), ‘Smart grid in Denmark 2.0: implementa-tion of three key recommendaimplementa-tions from the smart grid network’.
ENTSO-E (2014a), ‘ENTSO-E Network Code on Electricity Balancing’.
ENTSO-E (2014b), ‘ENTSO-E Target Methodology for Adequacy Assessment. Consultation material’.
Eurelectric (2014), ‘Flexibility and aggregation: Requirements for their interaction in the market’. Available online at: http://www.eurelectric.org/. Last accessed on 20/09/2017.
Fadlullah, Z. M., Duong, M. Q., Kato, N. and Stojmenovic, I. (2013), A Novel Game-based Demand Side Management Scheme for Smart Grid, in ‘IEEE Wireless Communications and Networking Conference (WCNC 2013)’, Shanghai.
GAMS (2018), ‘Bilevel Programs’. Available online at:
https://www.gams.com/latest/docs/UG EMP Bilevel.html. Last accessed on 21/06/2018.
Harbo, S. and Biegel, B. (2012), Contracting flexibility services, Technical report, iPower Consortium.
Hennings, W., Mischinger, S. and Linssen, J. (2013), ‘Utilization of excess wind power in electric vehicles’,Energy Policy 62, 139–144.
Katz, J. (2014), ‘Linking meters and markets: Roles and incentives to support a flexible demand side’, Utilities Policy 31, 139–144.
Kim, S. (2014), Game Theory Applications in Network Design, Information Science Refer-ence, Hershey.
Kitzing, L., Katz, J., Schr¨oder, S., Morthorst, P. and Møller Andersen, F. (2016), The residential electricity sector in Denmark: A description of current conditions, Working paper, Department of Management Engineering, Technical University of Denmark. The document is a working paper that has been written under the iPower research project.
Koponen, P., Ikaheimmo, J., Vicino, A., Agnetis, A., Pascale, G. D., Jimeno, J., Sanchez-Ubeda, E. F., Garcia-Gonzalez, P., Cossent, R. and Carames, N. R. (2012), ‘Toolbox for Aggregator of Flexible Demand’,Energy Conference and Exhibition (ENERGYCON), 2012 IEEE International pp. 623–628.
Lu, N. (2012), ‘An evaluation of the HVAC load potential for providing load balancing service’, IEEE Trans Smart Grid 3(3), 1263–1270.
Luo, Z. Q., Pang, J. S. and Ralph, D. (1996), ‘Mathematical programs with equilibrium constraints’. Cambrigde University Press.
Mohesian-Rad, A. H., Wong, V. W. S., Jatskevich, J., Schober, R. and Leon-Garcia, A.
(2010), ‘Autonomous Demand-Side Management Based on Game-Theoretic Energy Con-sumption Scheduling for the Future Smart Grid’, IEEE Transactions on Smart Grid 1(3), 320–331.
Nguyen, D. T. and Le, L. (2014), ‘Joint optimization of electric vehicle and home energy scheduling considering user comfort preference’,IEEE Trans. Smart Grid 5(1), 188–199.
O’Connell, N., Pinson, P., Madsen, H. and O’Malley, M. (2014), ‘Benefits and challenges of electrical demand response: A critical review’,Renewable and Sustainable Energy Reviews 39, 686–699.
Park, C. K., Kim, H. J. and Kim, Y. S. (2014), ‘A study of factors enhancing smart grid consumer engagement’, Energy Policy72, 211–218.
Roos, A., Ottesen, S. O. and Bolkesjo, T. F. (2014), ‘Modeling consumer flexibility of an aggregator participating in the wholesale power market and the regulation capacity market’, Energy Procedia 58, 79–86.
Saad, W., Han, Z., Poor, H. V. and Basar, T. (2012), ‘Game-Theoretic Methods for the Smart Grid: An Overview of Microgrid Systems, Demand-Side Management, and Smart Grid Communications’, IEEE Signal Processing Magazine 29(5), 86–105.
Short, J., Infield, D. and Freris, L. (2007), ‘Stabilization of grid frequency through dynamic demand control’, IEEE Transactions on Power Systems 22(3), 1284–1293.
von Stackelberg, H. (2011),Market Structure and Equilibrium, Springer, Berlin, Heidelberg.
Zugno, M., Morales, J., Pinson, P. and Madsen, H. (2013), ‘A bilevel model for electricity retailers’ participation in a demand response market environment’, Energy Economics 36, 182–197.
Appendix A Additional information about data
Figure A.1: Compensations to consumers and savings on payments for imbalance (Case 1)
Figure A.2: DK2 area hourly consumption in 2014
Figure A.3: Modelled HP load and actual outside temperature in 2014
Figure A.4: Modelled aggregator’s hourly imbalances
Appendix B Additional results
Figure B.1: Shifted consumption patterns during one week in winter (Case 2), Scenarios 5, 6 and 7
Figure B.2: Compensations to consumers and savings on payments for imbalance (Case 1)
Figure B.3: Differences in forecasted and actual reductions in total imbalance payments after optimisation (Case 1)
Figure B.4: Total imbalance cost before and after optimisation (Case 2) with a maximum of 20% error in the imbalance price forecast
Figure B.5: Total imbalance cost before and after optimisation (Case 2) when the aggregator takes the previous hour data for the imbalance price forecast
Chapter 2
Cooperative governance structures in flexible electricity
demand aggregation
Cooperative Governance Structures in Flexible Electricity Demand Aggregation
Per J. Agrell
∗, Ieva Linkeviciute
†September 9, 2018
Abstract
With an increasing share of renewable energy sources, the need for demand flexibility is growing. Large consumers have a potential to provide this flexibility to the market, however, in order to do that, they need right incentives and a favourable environment.
This paper examines whether the cooperative governance structures in flexible elec-tricity demand aggregation and trading could bring value to the market participants and final power consumers comparing to situations, where demand flexibility is traded individually or via the investor owned aggregator. We provided numerical estimation using Nord Pool intraday market data for DK2 price area in Denmark. We found that the cooperative of large consumers would offer the lowest price and the largest quantity of flexible demand in the long run. Moreover, sharing the fixed flexible de-mand coordination and market access cost would result in the highest profit for large consumers, giving the necessary incentives to stay in the market.
Keywords: demand-side management, flexibility, aggregation, electricity market, co-operative governance structure
JEL classification: C61, C63, C72, L94
∗Universite catholique de Louvain Louvain School of Management, Center of Operations Research and Econometrics (CORE), B-1348 Louvain-la-Neuve, Belgium. per.agrell@uclouvain.be
†Copenhagen Business School, Department of Economics, Porcelaenshaven 16 A, 1st floor, DK-2000 Frederiksberg, Denmark. ili.eco@cbs.dk
1 Introduction
Recent development in power systems resulting from green energy oriented policy and in-creasing share of renewable energy sources among power producers bring some new chal-lenges in sustaining power system stability and efficiency. However, advancing technologies create a favourable environment for the new market participants who were not able to pro-vide balancing services or adjust the consumption according to the system’s needs and trade their flexibility in the market.
The intraday market, which is a place to trade power closer to the real time and adjust traded volumes in the spot market according to the forecast corrections, will play a more important role in the future, since the production of intermittent renewable energy sources complicate the forecasting process. Demand-side management is more suitable for providing a short-term flexibility up to hours before the actual consumption (Linkenheil et al., 2017).
Thus, the intraday market has a great potential in accommodating demand flexibility of large consumers who want to trade flexibility on their own, because the reaction time for the load adjustments is relatively long, comparing to markets for regulating power.
The potential for available flexibility can be distinguished between installed capacity, theo-retical, technical, economical and achievable potential (Grein and Pehnt, 2011). Theoreti-cal potential is characterised by typiTheoreti-cal daily, weekly and annual available load variations.
Technical potential takes into account technical aspects of load shifting, while economical potential reflects only the cost-effective part of load shifting. Finally, achievable potential accounts for various barriers limiting the access to the flexible load, such as market require-ments, lack of knowledge and experience of the potential flexibility providers. Thus, the practically available flexible load is significantly lower compared to its installed capacity.
Demand-side flexibility providers can be divided into three groups: small consumers, such as households, large (commercial) consumers, such as supermarkets, hospitals, universities, and industrial consumers. In terms of theoretical potential, they differ in terms of their flexible load profiles. Usually, industrial consumers with energy-intensive processes are able to offer constant amounts of flexibility in each hour of the year, while households’ and large consumers’ flexible load varies depending on the season, weekday and hour during the day
(Grein and Pehnt, 2011). For example, residential demand-side management programmes are analysed based on active occupancy profiles by L´opez-Rodr´ıguez et al. (2013), and one of the factors that influence large consumers’ load is their working hours. The sources of flexibility for residential consumers include electric vehicles, washing machines, tumble dryers, dish washers, heat pumps and refrigerators; for large consumers the common source is cooling and heating activities.1 However, this paper does not consider the hourly, weekly or annual variability of flexible load and analyses a certain point in time, where the access to shiftable load is guaranteed from a technical perspective and limited only to economic incentives.
Demand-side flexibility and price elasticity of electricity demand are closely related. The elasticity of price depends on time and in a long term the demand is more elastic than in a short term. One explanation could be that in a long term consumers can adapt to price changes and change their consumption habits, appliances or use new technologies.
Also, consumers’ response to price changes vary between different consumers: in general, small consumers are less price elastic than large consumers. Price elasticity can also be different for different price ranges. For instance, when large industrial consumers face high prices, their price elasticity increases. Elasticity can be influenced by the type of tariff that consumers are charged. For example, when small consumers are charged using real time pricing, they tend to respond to prices less than when they are charged using a time-of-use tariff. One reason could be that consumers understand time-use-tariff better and, therefore, adjust their consumption.2 Thus, small and large consumers may need different compensation mechanisms to ensure their willingness to provide flexibility.
Even though small consumers have a potential for offering flexibility, their individual vol-umes are too low to place a bid at the market. Furthermore, the gain from this activity would be moderate considering required time and effort. Therefore, the aggregation of flex-ible demand and trading on behalf of small consumers became a widely discussed approach to deal with market access issues for this segment. European Energy Regulators state that
“there should be a requirement that all consumers have the opportunity to participate in
1For further discussion on demand response potential in Europe see Grein and Pehnt (2011).
2For more detailed discussion about price elasticity and electricity demand response see Risø National Laboratory, Ea Energy Analyses, RAM-løse edb report on demand response in Denmark (Andersen et al., 2006).
all relevant markets <...>” and they “recognise the benefits of introducing independent aggregation3 and propose that MSs [Member States] enable independent aggregation, un-less a national implementation assessment suggests an alternative that better serves sys-tem efficiency and can be implemented effectively” (Agency for the Cooperation of Energy Regulators (ACER) and Council of European Energy Regulators (CEER), 2017). Thus, independent aggregators are seen as a first choice for aggregating flexibility.
Large consumers, on the other hand, have the potential to trade in the market on their own, since they can meet the minimum bid requirements for some hours during the day.
Despite that, it is still not clear, what is the best strategy for the large consumer: to trade individually, access the market via the investor owned aggregator or form a cooperative of large consumers, where the members share part of the cost but can freely choose the volumes they want to deliver. Thus, this paper investigates whether cooperative governance structures have an advantage in aggregating and trading flexible electricity load at the intraday market comparing to cases where there is no coordination or where the coordinator is owned by investors.
The model is illustrated by providing a numerical estimation. Due to the complex expres-sions of the equilibrium outcomes, it is not easy to identify winners and losers in all analysed scenarios. By showing a numerical example representing the intraday market trading in one hour and market players’ participation cost in Nord Pool power market, we evaluate po-tential benefits and losses for different market players depending on the chosen governance structure. The input data is chosen carefully to reflect a real world trading in Nord Pool power market.
The rest of the paper is organised as follows. Section 2 provides a literature review. Section 3 presents the model, i.e. the coordination game and the players. Equilibrium analysis in every scenario is provided in section 4. In section 5, we show the numerical estimates of our model, also include the sensitivity analysis of the main input variables. Finally, in section 6, we make the conclusion and suggest possible directions for future research.
3According to the European Energy Regulators, “independent aggregator” is ”an aggregator that is not affiliated to a supplier or any other market participant”. (Agency for the Cooperation of Energy Regulators (ACER) and Council of European Energy Regulators (CEER), 2017)
2 Literature review
Intermediation and cooperation are concepts worth focusing on while analysing the emerging changes in energy markets. Intermediary can be defined as an entity, which is a link between different actors, such as producers and end-users (Grandcl´ement et al., 2015). However, the concept of intermediation has evolved with time. According to Saglietto (2017), commer-cial intermediation started with the commercommer-cial travellers in the Middle ages, then mafia intermediation emerged in the 1900, followed by financial intermediaries (banks) in 1970, logistic intermediaries (brokers) in 1980, economic (intermediary agents) in 1990, electronic (e-intermediary) in 2000 and finally cultural and legal intermediation in 2010. All these types of intermediation, however, have the same function – they all coordinate and control physical, financial, informational or cultural flows, processes and activities (Saglietto, 2017).
A wide range of intermediary types result in a large number of studies on intermediation.
An interested reader is referred to Appendix A for a more detailed intermediation literature review. This study focuses on intermediation in power markets, where the aggregator acts as a link between large consumers and the intraday market.
Among many services provided by intermediaries4, the reduction of transaction cost com-pared to the situation where the parties interact directly, can increase social welfare. The concept of transaction cost was introduced by Coase (1937) as “a cost of using a price mech-anism”. This cost includes search and information, bargaining and decision, and policing and reinforcement costs. According to Coase, transaction cost can help to explain the emer-gence of firms. Transaction cost analysis became a more popular topic with Williamson’s work. Williamson studied “the comparative costs of planning, adapting, and monitoring task completion under alternative governance structures” and came to a conclusion that “gover-nance structures that have better transactional cost economizing properties will eventually displace those that have worse”(Williamson, 1981). He also claimed that there are two sides of vertical integration: on the one hand, when the transactions are organised within a firm, decision rights are centralised and this reduces bargaining cost and the risk related to bargaining impasse; on the other hand, the executives may extract rents in inefficient ways.
4See Saglietto (2017)